Journal dNatural Products Vol. 5 1 , NO.4 , pp. 690-702,JuI-Au~1988
690
MACROCYCLIC PYRROLIZIDINE ALKALOIDS FROM SENECZO ANONYMUS. SEPARATION OF A COMPLEX ALKALOID EXTRACT USING DROPLET COUNTERCURRENT CHROMATOGRAPHY' LEONH. ZALKOW,*CLARITAF. ASIBAL,JANA. GLINSKI, SANDRAJ. BONEITI, LESLIET . GELBAUM,~ DONALDVANDERVEER,and GARTHPOWIS'
Schwl of Chemistry, Georgia lnstitnte of Technology, Atlanta, Georgia 30332 AssTucr.-Ten 12-membered macrocyclic pyrrolizidine alkaloids, all of them esters of the necines, retronecine or otonecine, have been isolated from Serzerio anonymus. T h e separation, carried out by droplet counter-current chromatography, afforded senecionine [l],integerrimine [2], retrorsine 131, senkirkine [5], neosenkirkine 161,otosenine [lo], hydroxysenkirkine and a new alkaloid given the trivial name anonamine [9].Traces of usaramine 141 and another new alkaloid, hydroxyneosenkirkine [SI,were detected by 'H nmr. In addition, the previously unreported 3a~-hydroxy-4-ethoxy-2,6-perhydroindoledione [ll]was isolated. Xray structures were obtained for neosenkirkine [6], hydroxysenkirkine anonamine [9],and 111). 'H-''C heteronuclear shift correlated nmr (HETCOR) provided unambiguous chemical shift assignments for '3C-nmr data. Antitumor activity was assayed using the A204-rhabdomyosarcoma cell line in soft agarose.
m,
En,
Of the pyrrolizidine alkaloid (PA) containing plants, the genus Senecio (Compositae), having the largest number of species (ca. 1450), has generated numerous studies. The most current lists (1-3) of investigated species, however, indicate that only about 10% of the genus has thus far been studied. Most of these contain hepatotoxic PAS(esters of 1,2- unsaturated necines), and a number of these hepatotoxic alkaloids have been shown to be mutagenic. Pyrrolitidine alkaloids and their pyrrolic metabolites have been implicated in megalocytosis and mitotic inhibition (l),and recently semisynthetic pyrrolizidine alkaloid N-oxides have been investigated as antitumor agents (4,5). The work described in this paper, which continues our earlier work (6) on isolation and structure elucidation of pyrrolizidine alkaloids, involves the separation of pyrrolizidine alkaloids from a locally abundant species, Senecio anonymus Wood (formerly called Senecio smullii) from which a cytotoxic compound, jacaranone ethyl ester, was previously isolated in our laboratory (7). RESULTS AND DISCUSSION Examination of the alkaloidal fraction from S. anonymas led to the identification of ten 12-membered macrocyclic pyrrolizidine alkaloids. Four of these, senecionine 117 (8,9), integerrimine 121(9, lo), retrorsine 131 (10-12), and usaramine 147,known also as mucronatinine (10,13), are esters of retronecine while the six remaining, senkirkine 151 (14-16), neosenkirkine 167 (17,18), hydroxysenkirkine 17'1 (15,19), hydroxyneosenkirkine 187, anonamine 191, and otosenine 1107 (20,2 l),are esters of otonecine. The macrocyclic ester rings are formed by six different but closely related necic dicarboxylic acids. Thus, 1 and 5 are esters of senecic acid, 2 and 6 are esters of integerrinecic acid, 3 and 7 are esters of isatinic acid, and 4 and 8 are esters of trans-isatinic acid, while 10 is an ester of jacobinic acid and 9 is an ester of 7-hydroxyintegerrinecic acid. Since senecic and integerrinecic acids as well as isatinic and trans-isatinic acids are
'Taken from the Ph.D. dissertation of Clarita Florendo Asibal, School of Chemistry, Georgia Insritute of Technology, Atlanta, Georgia 30332, June, 1987. 2Research Center for Biotechnology, Georgia Institute of Technology, Atlanta, Georgia 30332. 'Department of Oncology, Mayo Clinic and Foundation, Rochester, Minnesota 55905.
Jul-Aug 19881
Zalkow etal. : Pyrrolizidine Alkaloids
69 1
geometric isomers, the eight esters 1-8 constitute four pairs of alkaloids differing only in the configuration of the C- 1 5 4 - 2 0 double bond. Therefore, the physical properties of these geometric isomers are very similar, posing a serious problem in separation. The initial EtOH plant extract was partitioned to give the N-oxides in an H,O layer and the free bases in a CHCI, layer. The ratio of N-oxides to free bases varied from 6: 1 to 10:1 depending on the plant parts (leaves, flowers, stems). Separate analyses of the extracts from the flowers, leaves, and stems combined with roots revealed a specificdistribution of the alkaloids. Thus, the major PAS from the flowers were 10, 1, and 3, from the leaves 5 , 6 , and 7, and from the stems and roots 5,6,3, and 1. The total estimated alkaloid content of the whole plant was 0.02%, and the relative percentages of the various alkaloids were 5 (41.0), 6 (20.5), 1 ( 1 3 3 , 7 (7.8), 10 (6.7), 3 (4. I), 9 (3.8), 2 (2.0), 8 (0.5),and 4 (0.3%).
1 R1=R3=Me, R2=H (senecionine) 2 R1=R2=Me, R 3 = H (integerrirnine) 3 R'=CH,OH, R2=H, R3=Me (retrorsine) 4 R 1 = C H 2 0 H , R2=Me, R3=H (usamnine)
M'e
5 R'=R3=Me, R2=H (senkirkine) 6 R'=R2=Me, R 3 = H (neosenkirkine) 7 R ' = C H 2 0 H , R2=H, R3=Me (hydroxysenkirkine) 8 R'=CH20H, R2=Me, R 3 = H (hydroxyneosenkirkine)
9 R'=Me, R2=CH20H, R 3 = H (anonarnine) 10 R1=R3=Me, R2=H, epoxide (15S, 20s) in place of AI5 (otosenine)
Several chromatographic methods, including gravity column, tlc, centrifugal tlc, hplc, and droplet counter-current chromatography (dccc) were evaluated for their efficiency in the separation of the complex alkaloid mixture. Finally, dccc was chosen as the most suitable method for preparative separation. As we have observed many times, PAS tend to adsorb irreversibly to solid phases such as alumina, silica, or reversed-phase in
692
Journal of Natural Products
Wol. 5 1, No. 4
hplc, resulting in large losses. DCCC,a technique based on partition between two immiscible liquid phases, is free of this drawback. Moreover, with a properly selected solvent system, it gives good separation with low solvent consumption (separation of 3 g of a crude mixture consumed only about 4 liters of solvent mixture). In an attempt to find a solvent system most suitable for our needs, we measured the partition coefficients of the model compounds 9-benzoylretronecine, indicine, monocrotaline, and retronecine in the three solvent systems containing CHCI3-MeOH-H2O in ratios of 13:7:4, 7: 13:8, and 5:6:4 and in CHC13-C6H6-MeOH-H20,5:5:7:2 (22). This group of model compounds provides a representation of low, medium, and high polarity PAS. The first three solvent systems gave fairly similar results and were suitable for separation of more polar PAS, while the C6H6-containing system was found to be better suited for moderately polar PAS. Because most of the alkaloids present in S. anonymas are moderately polar, this system was selected for the initial run. The collected fractions were monitored by tlc and by 300-MHz 'H nmr. The 'Hnmr spectra were especially useful in determining compositions of the fractions containing mixtures of cis-trans isomeric alkaloids; these could be distinguished by their H20 absorptions which occur at 6 6.5-6.8 for the cis isomers 2,4,6, and 8 and about 6 5.7-5.9 for the trans isomers 1 , 3,5,and 7 . Overall results of the dccc separation exceeded our expectation, affording pure samples of 8 out of a total of 10 alkaloids present in the plant plus the non-alkaloidal compound 1 1 . Only two very minor components, 4 and 8, have not been fully separated from their stereoisomers 3 and 7,respectively; their 'H-nmr data were acquired from the most enriched fractions. Attempts to separate the pairs of cis-tram diastereomers using other chromatographic methods (gravity columns, traditional and centrifugal tlc, and reversed-phase hplc) resulted only in slight enrichments, low recoveries, andor broad peaks. The stereoisomers 1-2,3-4, 5-6,and 7-8differ only in the configuration about the C- 15-C-20 double bond. 'H-nmr spectroscopy readily distinguishes between stereoisomeric pairs by the chemical shift positions of the H-20 quartet which appears at 6 5.70-5.86for thetransisomersl, 3,5,and7andat66.49-6.77forthecisisomers2, 4,6,and 8 (see Table 1). In the new alkaloid, anonamine 191, instead of a quartet, an AB pattern is seen at 6 6.63 for H-20. This signal was shown by both decoupling experiment and 'H- 'H shift correlated spectroscopy(COSY; Figure 1) to be coupled to the two sets ofdoublet-of-doublets at 6 4.19 and 4.40. Anonamine 191differs from neosenkirkine 161 only by the replacement of the C-21 methyl with a hydroxymethylene group. The postulated structure of another new alkaloid which we have named hydroxyneosenkirkine E81 is based on a comparison of its 'H-nmr spectrum with those of hydroxysenkirkine 1 7 and neosenkirkine 161.The 'H-nmr spectra of 7 and 8 are very similar, displaying characteristic and essentially identical AB patterns for the H,-18 at 6 3.66 and 3.74 in 7 and at 6 3.69 and 3.76 in 8. The two spectra differ significantly only in the areas of olefinic absorption. Isomer 8 also resembles 6,where the H-20 absorption is found at 6 6.7 1. The structures of 6,7; and 9 were confirmed by single crystal X-ray crystallographic analyses (23). Structures of the other alkaloids were established by melting points, optical rotations, high resolution ms and m, and direct comparisons with authentic samples (1, 3,10)andor with literature data (2,4,5). 13C-nmrspectra were obtained for all the alkaloids except 4 and 8 (see Table 2), and unambiguous assignments for all H-containing carbons were based on 'H-13C heteronuclear shift correlated n m r (HETCOR; Figure 2). For example, the clear distinction between the chemical shifts of c-2 and c-20 in retrorsine 131,which were previously interchanged (24) is illustrated in Figure 3. Table 2 indicates chemical shifts that were misassigned or interchanged in earlier reports. As seen in Table 2, cis isomers 2,6,and 9 show more shielded C- 14's with values of
.
69 3
Zalkow et al. : Pyrrolizidine Alkaloids
h
I
VI
m ?
N
C Y
e
a
. . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
[Vol. 5 1, No. 4
Journal of Natural Products
694
1
-
a
0
4
. . 0
i
a0 ma 8 I
u w
0
.
a
01
0
0 .
a
a
8
7
6
I
I
5
4
.-
T-
3
2
1
F 1 (PPM) FIGURE 1.
'H-'H shift correlated spectroscopy (COSY)of anonamine 191.
0
Zalkow et al. : PyrrolizidineAlkaloids
Jul-Aug 1988]
C
3
. . . . . .
. . . . . . - m
. . . . . . r
. . . . . . . . . . . . . . . . . . n n
. . . . . . ~
. . . . . . . . . . . . . . . . . . ~ w
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . n - - ~
. . . . . . z
695
. . . . . . . . . . . . . . . . . . q r z ~ z ~ ; ~
rw.5 1, NO.4
Journal ofNatural Products
696
L
I
1
5.5
I
I 4. 5
'
I
3.5
I I1 I'
2.5
I I
1.5
Oa5
I &
'I
18016014012010080 6 0 40 2 0 0
F 2 (PPM) FIGURE 2.
*H-l3Cheteronuclear shift correlated nmr (HETCOR)of retrorsine 133.
29.68,28.87,and 30.34ppm, respectively, because ofthe cis interaction with the C2 1 groups, while corresponding values for the trans isomers 1, 3, 5 , 7, and 10 are 38.23,34.87,37.60,37.17,and 35.37,respectively. ThechangeoftheC-18methyl group in 5 to a hydroxymethylene group in 7 is evident from the downfield shift of this carbon from 24.47 to 66.30ppm. In a similar manner, the change from a C-2 1 methyl group in 6 to hydroxymethylene group in 9 results in a chemical shift change from 14.46 to 59.28 ppm. There are a number of reports on the co-occurrence of the diasteromers 1-2 (11,16,25), 3-4 (9,25), and 5-6 (17,18).Fot the pair 7-8, this is the first report of isomer 8 and only the third time that 7 has been reported (15,191. On the basis of the literature cited above, it seemed to be an unusual coincidence to find 4 pairs ofgeometric isomers in one plant. Therefore, we attempted to examine the possibility of cis-trans interconversion by subjecting the N-oxide of 1 to the same work-up as that used for the crude extract, including zinc-acid reduction. The test revealed no formation of the isomeric 2. Cis-trans isomerization has been induced by uv irradiation (25)to convert 4 into 3 and by brominatioddebromination ( 2 6 ) to convert 3 into 4 . The identity of a new, non-alkaloidal compound, isolated from the crude alkaloidal extract, was established as 3a-hydroxy-4-ethoxy-2,6-perhydroindoledione Ell}by X-
Jul-Aug l988]
Zalkow eta!. : Pyrrolizidine Alkaloids
697
6. 2
6. 0
5.8
5. 6 146
142
138
134
130
126
122
F 2 (PPM) FIGURE 3.
Expanded porrion of ‘H-’.k HETCOR nmr of retronine [3].
+ +
ray analysis.* A quadrant of data (? h, k, 1) was collected on a Syntex P2 diffractometer using omega scans. A total of 1884 unique data were measured out to 28 = 50°, and 1197 reflections were used in a full-matrix least-squares refinement on F. The refinement converged at R=0.072, and R w z 0 . 6 8 for 151 parameters varied. Most of the hydrogens were located from a difference Fourier map and were included in the refinement at fixed positions. The hydrogens on C-10 and C-11 were calculated as members of fixed groups. Figure 4 shows a computer-drawn picture of 11.The compound bears close resemblance to a number of natural products isolated recently from algae (27,28), but contrary to these, it contains a trans-fused 2,6-perhydroindoledione skeleton. The monoclinic crystal belonging to the space group P2 ,/n showed unit cell parameters a=5.6772(8), b=26.831(4), c=7.174(2) A, p = 101.00(2)0, Z = 4 , D, = 1.317 g ~ m - X~ =, 0.7 1969 A and has a center of inversion, indicating that the crystal was racemic. Severe overlap of the nmr signals even at 400 MHz made it necessary to determine the chemical shifts of the individual protons using a COSY experiment. Analysis of the COSY data indicated that H-3a1,H-5a, and H-7a overlapped at
4Atornic coordinates for this structure have been deposited with the Cambridge Crystallographic Data Centre and can be obtained on request from Dr. Olga Kennard, University Chemical Laboratory, Lnsfield Road, Cambridge, CB2 lEW, UK.
mol. 5 1, No. 4
Journal of Natural Products
698
02
FIGURE 4. Computer-drawn picture of compound 11.
6 2.83 and in addition, H-5P and H-7P overlapped at 6 2.53. AI1 chemical shift and coupling constant data are consistent with the structure determined from the X-ray analysis. The origin of 11 remains unclear; most likely it is not a natural product but rather an artifact arising from jacaranone {121,a known constituent ofS. anonymus (7), during the work-up, which included the use of ErOH and NH,. Reexamination of the crude plant extract, avoiding the use of NH, and EtOH, did not lead to the detection of 11. Alkaloidal fractions frequently are found to contain neutral compounds carried along in the routine work-ups of plant material. The relative in vitro cytotoxicities of some of the isolated alkaloids and related compounds were measured against the A204 human rhabdomyosarcomaceJ1 line using the soft agar colony forming assay (Table 3). Indicine N-oxide was used as an internal standard for each group of compounds in Table 3 , because it is a pyrrolitidine alkaloid that was selected for clinical development as an antitumor agent by the National Cancer Institute (29). Compounds 1,2,3,9, and 10 show similar cytotoxicity, while 5 , 6 , and 7 are less cytotoxic, but all of the compounds tested were more cytotoxic than indicine N-oxide in this system. Senecionine N-oxide exhibited the same cytotoxicity as senecionine. In the bottom of Table 3 (last 3 entries) monocrotaline 1141, the semisynthetic compound 15,and their N-oxides are compared with indicine E131 and
11
12
Jul-Aug 19881
699
Zalkow et al. : Pyrrolizidine Alkaloids
=g
13
14
15
its N-oxide. Semisynthetic 15 N-oxide shows excellent cytotoxicity in this assay and excellent in vivo activity (5). We have found an excellent correlation between cytotoxicity against the A204 rhabdomyosarcoma cell line and in vivo activity for a large number of semisynthetic compounds such as 15, and in every case the N-oxides were more cytotoxic than the free bases.5 The semisynthetic bases were not screened in vivo because of their expected hepatotoxicity. The natural alkaloids indicine f13] and monocrotaline 1141 are more cytotoxic than their corresponding N-oxides, but neither monocrotaline nor its N-oxide shows much activity in this test. None of the naturally occurring macrocyclic alkaloids reported in this study would be expected to be useful antitumor agents because their hepatotoxicity would severely limit their usefulness (1). EXPERIMENTAL GENERAL EXPERIMENTAL PROCEDURES-AII separations were carried out using a Buchi 670 dcc Chromatograph equipped with 300 tubes of 2.7 mm i.d. plus 200 tubes of 3.0 mm i.d. The flow rates varied between 15 and 24 mYh, and eluates were collected by an automatic fraction collector. The separation process was monitored by a combination of tlc analysis on EM A,O, 150 F-254 plates developed in a mixture of toluene with 5 to 15% MeOH or in CHCI, with 5 to 10% MeOH and 'H-nmr analysis. All 'H- and I3C-nmr spectra were obtained using a Varian XL-400 spectrometer operating at 399.934 MHz and 100.575 MHz, respectively. Chemical shifts are reported relative to residual CHCI, (7.24 ppm) for 'H and to CDCI, (77.0 ppm) for 13C. 'H-13C heteronuclear shift correlated nmr spectra were collected as a 128 X 4096 data matrix using the pulse sequence HETCOR supplied with the Varian 6.1 c software (30,31 ) . This was processed using pseudo echo weighting to a 5 12 X 2048 data matrix for plotting. The hplc experiments were performed on a L D C Constametric 111pump equipped with a Rheodyne 7 120 sample injector and either a Universil C,,, 25 cm X 4.6 mm or an W T E C H C,,, lop, 25 cm X 10 mm column and a Holochrome Gilson uv detector (at 220 nm). Two solvent systems were employed: 20-35% EtOH in 0.01 M(NH,),CO, and l0-50% MeCN inO.O1 M(NH,),CO3(pH 7 . 6 ) .Optical rotations were taken with a Perkin-Elmer 141 polarimeter. Mass spectra were obtained on a Varian MAT 112s spectrometer interfaced with an SS200 data system. Melting points were taken on a Kofler hot stage and are corrected.
5L.H. Zalkow, unpublished results.
Journal of Natural Products
700
{Vol. 5 1, No. 4
TABLE3. Cytotoxicity against A204 Rhabdomyosarcoma Cell Line In Vitro.' Compound
1 . . 2 . . 3 . . 5 . .
6 . 7 . 9 . 10 . 13 . 14 . 15 .
. . . . . .
150?6
15026
. . . . . . 12055 . . . . . . 12025 . . . . . . 260220
. . . . . . . 221+13
. . . .
. . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
360+50 153+4 130+9
13 . . . . . . . .
34?11
3 16*95
>loo
440220 721238 1120 125222
%e Experimental section for details of preparation of soft agarose cultures. Cultures were conducted in quadruplicate to allow reliable estimates of the variance of the IC,, to be obtained. Control cultures with vehicle alone were always run at the same time. Dose-response curves were constructed using at least four drug concenrrations to produce between 10 and 99.9% inhibition of cell growth. Dose-response curves were constructed on at least three different preparations. 90obtain the IC,,, the drug concentration producing 50% inhibition of cell growth, and its variance, the dose-response data was fitted to a monoexponential curve using a NONLIN nonlinear least squares regression analysis program. Variance of IC,, was obtained from the variance of the intercept and slope using Taylor series expansion. Values are mean ? SE.
IN VITRO curoToxicin.--Soft agarose cultures of A204 human rhabdomyosarcoma cells were performed as follows: Each 35-mm culture dish contained a base layer consisting of 0.5 ml Dulbecco's modified Eagle's medium containing 10% fetal calf serum with 0.5% agarose (growth medium). On day 0 cells in bulk culture were dissociated with trypsin and EDTA, washed once in growth medium, and tubcultured by layering 1 X 10' viable cells in 0.5 ml growth medium with 0.3% agarose over each base layer. Cultures were examined with the aid of an inverted stage microscope, and only cultures containing uniformly distributed single cell suspensions (< 10 30-pcell cultures and no 60-p clusters) were accepted for subsequent evaluation. Cultures were maintained in cell culture incubators at 37', 5% CO,, 95R air, and 100% humidity. O n day 1 (24 h later) an upper layer of 1 ml growth medium with and without the compound under investigation was added to the dishes. After 24 h, the upper layer of medium was removed by aspiration, agarose culture surfaces washed once with 0.5 ml prewarmed growth medium, and then overlaid with 1 ml offresh growth medium. Colony formation was examined at daily intervals by conventional light microscopy. Cell lines form a sufficient number of detectable colonies (>60-pdiameter) for analysis following 7 to 9 days incubation. Viable colonies were stained using a metabolizable tetrazolium salt (2-piodophenyl-3-p-nitrophenyl-5-phenyltetrazolium chloride), and colonies were counted using a Bausch Si Lomb FAS-I1 image analysis system. Cultures were conducted in quadruplicate. Control cultures without drug were run at the same time. ISOLATION OF THE ALKALOIDAL FRACTION.-Flowering s. anrM]u///Swas collected 10 miles south of Atlanta, Georgia, in the third week of May 1984, and identified by Dr. Caywood Chapman, Department ofkience, Gordon Junior College; a voucher specimen is on deposit at the Georgia Institute ofTechnology. Different parts of the plant-flowers, leaves, and stems + roots-were dried separately. The air-
Jul-Aug 19881
Zalkow et al. :
70 1
PyrrolizidineAlkaloids
dried flowers (3.3 kg) were macerated in a blender with 95% EtOH and allowed to soak in about 5 gal of solvent for at least 2 days at room temperature. The solvent was replaced with fresh solvent three times, and the combined extract was evaporated on a rotary evaporator to leave 0.7 kg of a dark residue. Out of this, 333 g was partitioned between H,O and CHCI, (2.0 liters each). The organic layer gave 60.6 g of material which was partitioned between 0.25 liter each of hexane and 90% aqueous MeOH. From the aqueous MeOH layer 15.O g ofa residue was obtained. This was dissolved in 200 ml of 5% NaOH and extracted five times with 300 ml of Et,O. The Et,O extract was then washed three times with 150 ml of 10% HCI. Basification of the combined aqueous layer with excess of N H 4 0 H , followed by extraction with CHCI,, drying over MgS04, and evaporation in vacuo, yielded 106 mg ofa crude alkaloidal fraction (0.007% dry wt). The aqueous layer, 2.0 liters, from the initial CHC1,-H,O partition was treated with 120 ml of concentrated H,SO, and 24 g of Zn powder. The reaction mixture was stirred overnight and the decanted so lution extracted with 4 X 400 ml ofCHCI,. The aqueous phase was then made alkaline with excess ofaqueous N H and extracted four times with 400 ml of CHCl;. The combined CHCI, extract was dried over MgS04, filtered, and evaporated in vacuo leaving 1.28 g ofa crude alkaloid fraction (0.04% dry w). The air-dried leaves, 3.37 kg, were processed in the same manner as the flowers, giving 930 g of a concentrated EtOH extract. The workup afforded 187 mg of the free alkaloid fraction plus 1.98g after Zn reduction of the N-oxides (0.006% and 0.06% dry wt, respectively). Due to a very low content of PA in the stalks and roots, the EtOH extract (3 11 g) obtained from 3.0 kg of the dry material was subjected to Zn-acid reduction without prior separation ofthe free alkaloids. The work-up gave 372 mg of the alkaloid mixture (0.01% dry wt). Combined leaves and flowers (4.3 kg), collected in the third week ofMay 19'8s in the same vicinity as previously (voucher specimen at Georgia Institute ofTechnology), were extracted, as before, to give 879 g of concentrated EtOH extract which was directly reduced to give 4.67 g of crude alkaloid fraction (0.01Q dry wt). The dcc chromatograph was filled with the stationary phase consisting of the lower layer of a solvent mixture prepared from CHCl,-C,H6-MeOH-H,0 (5:5:7:2). The alkaloidal fraction of S. anonymus (1985 collection), 3.0 g , dissolved in 15 ml ofa 1: 1 mixture of the upper and lower phases was aspirated into the instrument followed by the mobile phase (upper layer). The fractions, of 20-1111 vol, afforded the following alkaloids: 23-26, 9 (37 mg); 27-29, 11(30 mg); 30, 7 (17 mg); 3 1-34, 7 8 (64mg); 3 7 4 3 , 10 (65 mg); 5 8 - 6 0 , 3 + 4 ( 1 3 m g ) ; 61-64,3(30mg);6849,5(29mg); 70-77,5+6(556mg);78-84,6(15 mg). From fraction 87 on, the stationary phase started to be pumped out and collected, giving: 114-1 15, l ( 2 0 . 5 m g ) ; 116-119, 1 + 2 ( 1 2 1 . 5 m g ) , 1 2 0 , 2 ( 8 m g ) .
+
SEPARATIONOF SENKIRKINE 151 AND NEOSENKIRKINE [6].-A mixture containing 5 and 6 (ca. 1: 1) (158.7 mg) was subjected to dccc in the solvent system CHCI,-C6H6-MeOH-H,0 (5:5:7:2) in ascending mode. At the Bow rate of 15 mUh, fractions of 10 ml were collected. Pure 5 (24.8 mg) was eluted in fractions 133-146 while pure 6 ( 5 . 7 mg) was obtained from fractions 176-182. Fractions 147-175 contained mixtures of varying ratios of 5 and 6 . SEPARATION OF HYDROXYSENKIRKINE
(7AND HYDROXYNEOSENKIRKlNE [8].-A
mixture Of
7 and 8 (ca. 9: 1) (94 mg) was subjected to chromatography in the solvent system CHCl3-MeOH-H,O (13:7:4) in descending mode. At Bow rate of 24 mUh, 20-1111 fractions were collected. Fractions 36-37 afforded a mixture of 7 and 8 (9.5 mg) in a 1:2 ratio. Pure 7 (44.1 mg) was obtained from fractions 43-49. Intervening fractions contained varying amounts of both isomers.
ATTEMPTEDSEPARATION OF RETRORSINE [3] A N D USARAMINE [4].-A mixture of 3 and 4 (93:7) (40 mg) was subjected to dccc using the solvent system CHCI,-MeOH-H,O (13:7:4) in descending mode. At flow rate of 24 mUh, 20411 fractions were collected. Fraction 14 (20.5 mg) afforded 3 with traces of 4,and fractions 15-17 (16 mg) afforded pure retronine. Fraction 14 was rerun under the same conditions, but no pure usaramine was obtained. Pure 3 (16 mg) was obtained in fractions 17-22. SEPARATION OF SENECIONINE [I]AND INTEGERRIMINE [21.-A mixture of 1and 2 (1: 1) was subjected to fractional crystallization from Me,CO utilizing the fact that 1was less soluble than 2 in the solvent.
ANONAMINE[91.-Mp 202', [UI2'D +33.5" (c= 1.0, CHCI,); eims m h (%) 100 (15), 110 (46), 122 (259, 123 (49), 124 (159, 150 (16), 15 1 (63), 168 (50), 169 (24), 248 (21), 266 (13), 282 (23), [MI+ 381 (2); cims [MH]+ 382 (100); exact mass calcd for C,9H,7N07, 381.1788, found 381.1742. 3a~-HYDROXY-4-ETHOXY-2,6-PERHYDRoINDOLEDlONE Ill].-Mp 170'; eims m/z (%) 43 (loo), 44(42), 55(24), 70(28), 71(23), 73(24), 97(16),99(39), 125(37), 141(40), 150(36), 167(25), 184 (2), 195 (5), 213 (0.8), 214 (0.6); cims (MHf+ 214 (100); exact mass calcd for C,,HI4NO4 CMH}+ 214.1185, found bycims214.1048; 'Hnmr(CDC1,)G l . l ( t , ] = 7 Hz, CH,Me), 3.60and3.35(each d q J z 9 . 5 , 7.0Hz,CH2Me),4.04(dd,]= 13.7,4.9Hz,H-4), 3.72(dd,]=4.7, 1.7Hz,H-7a), 2.83
Journal of Natural Products
702
and 2.21 (eachd,]= l5.0Hz, H,-3), 2.81 (m, H-5), 2.52(d,]= (d,]= 13.8 Hz, H-7).
[Vol. 5 1, No. 4
15.5 Hz, H-5), 2.85 (m, H-7), 2.53
ACKNOWLEDGMENTS W e thank the National Cancer Institute, National Institutes of Health (CA 3 1490) for partial support of this work. C.F.A. is indebted to the Institute of International Education for a Fulbright grant in partial support of this work. We express our sincere appreciation to Dr. C.C.J. Culvenor, CSIRO, Australia, for authentic samples of otosenine, retronine, and senecionine. LITERATURE CITED 1. 2. 3. 4.
5. 6. 7.
8.
9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.
A. R. Mattocks, “Chemistry and Toxicology of Pyrrolizidine Alkaloids,” Academic Press, Orlando, Florida, 1986, pp. 18-26. L. W . Smith and C.C.J. Culvenor,]. Nut. Prod., 44,129 (198 1). D.J. Robins, Fmtscbr. Chem. Org. Naturst., 41, 115 (1982). L.H.Zalkow, J.A. Glinski, L.T. Gelbaum,T.J. Fleischmann, L.S. McGowan, and M.M. Gordon, J . Med. C h . , 28, 687 (1985). L.T. Gelbaum, M.M. Gordon, M. Miles, and L.H. Zalkow,]. Org. Chem., 47, 2501 (1982). L.H. Zalkow, S . Bonetti, L.T. Gelbaum, M.M. Gordon, B.B. P a d , A. Shani, and D. VanDerVeer, J . Nut. Prod., 42, 603 (1979). L.T. Gelbaum, L.H. Zalkow, and D. Hamilton,]. Nut. Prod., 45, 370 (1982). X.T. Liang and E. Roeder, Planta Med., 50, 362 (1984). M.J. Pestchanker, M.S. Ascheri, and O.S.Giordano, PlantaMed., 2, 165 (1985). R.S. Sawhney and C.K. Atal, J . Indian C h m . Sor., 47, 667 (1970). J.N. Roitman, A u f t . ] . Chem., 36,1203 (1983). Y.Asada, T. Furuya, T . Takeuchi, and Y . Osawa, Planta Med. , 46, 125 (1982). N.S. Bhacca and R.K. Sharma, Tetrahedron, 24, 63 19 (1968). L.H. Briggs, R.C. Cambie, B.J. Candy, G.M. ODonovan, R.H. Russell, and R.N. Seelye,]. Org. Chem., 2492 (1965). D.H.G. Crout,]. Chem. Soc., Perkin Tram. I , 1602 (1972). D.S. Bhakuni and S . Gupta, Planta Med., 46,25 1 (1982). F.M. Panizo and B. Rodriguez, An. Quim., 70, 1043 (1974). Y. Asada and T. Furuya, Planta Med., 44,182 (1982). F. Bohlmann, C. Zdero, J. Jakupovic, M. Grenz, V. Castro, R.M. King, H. Robinson, and L.P.D. Vincent, Phytochemistry, 25, 115 1 (1986). E. Roder, H . Wiedenfeld, and A. Hoenig, Planta Med., 49, 57 (1983). J.F. Resch, S.A. Goldstein, and J. Meinwald, Planfa Med., 49, 255 (1983). H . Otsuka, Y.Ogihara, and S. Shibata, Phytochemistry, 13, 2016 (1974). J.A. Glinski, C.F. Asibal, D. VanDerveer, and L.H. Zalkow, Acta Crystallogr., Sat. C, in press. S.E. Drewes, I. Antonowitz, P. b y e , and P.C. Coleman, J . Chem. Sor., Perkin Tram. I , 287 (198 1). C.C.J. Culvenor and L.W. Smith, A w t . ] . Chem., 19, 2127 (1966). A.R. Mattocks,]. C h . Soc. C, 3, 225 (1968). M. D’Ambrosio, A. Guerriero, and F. Pietra, Helv. Chim. Acta, 67, 1484 (1984). A. Guerriero, M. D’Ambrosio, P. Traldi, and F. Pietra, Natunuissenschajen, 71, 425 (1984). S.A. King, M. Suffness, B. Leyland-Jones, D.F. Hoth, and P.J. ODwyer, Cancer Treat. RQ., 71, 5 17 (1987). A. Bax,J . M a p . Reson., 53, 517 (1983). J.A. Wilde and P.H. Bolton,]. M a p . Reson., 59, 343 (1984). R.J. Molyneu, J.N. Roitman, M. Benson, and R.E. Lunden, Phytochemistry, 21, 439 (1982). N.V. Mody, R.S. Sawhney, and S.W . Pelletier,]. Nut. Prod., 42, 4 17 ( 1979). A.J. Jones, C.C.J. Culvenor, and L. W . Smith, A u s t . ] . Chem., 35, 1173 (1982).
Rem;wd 23 Nowmber 1987
